La Jornada: UN: uncontrolled advance of AI could pose catastrophic risks

Geneva. A group of independent United Nations experts warned yesterday that advances in artificial intelligence (AI) are outpacing scientific knowledge and government policies, meaning there are no guarantees the technology will not cause catastrophic harm.

A preliminary report from the UN Independent International Scientific Panel on Artificial Intelligence indicated that policymakers face a growing dilemma: They need robust evidence to regulate AI effectively, but such evidence is failing to keep pace with the technology’s rapid evolution.

“AI capabilities are surpassing both scientific knowledge and the ability of governments to adapt,” said Yoshua Bengio, co-chair of the panel, made up of 40 experts from different regions.

“In the face of mounting evidence of deceptive AI behavior, science cannot currently guarantee that, as its capabilities continue to increase, AI will not cause catastrophic harm, either by itself or by malicious users,” he added.

Described as the first independent global assessment of the risks and opportunities of AI, the report aims to provide up-to-date assessments of the science to help guide decision-making as governments grapple with rapidly evolving systems.

In the near term, a shift towards “agent” AI systems capable of carrying out real-world tasks is expected, although growth could be limited by shortages of energy and high-quality data.

Over time, AI is expected to be self-improving, more deeply integrated into the economy and in convergence with technologies such as quantum computing and biotechnology.

AI already demonstrates expert-level reasoning in mathematics and science and is accelerating the development of drugs and vaccines. Additionally, the complexity of their tasks doubles every four to seven months, which could allow systems to complete jobs that take humans days or weeks, according to the report.

Although this could bring significant economic benefits, it is not yet clear whether the productivity gains from the use of AI will translate into more widespread growth or affect employment.

The think tank also flagged a number of security concerns, such as the risk of losing control over AI systems as they become increasingly autonomous and deceptive.

AI is already being used to generate disinformation and other harmful content, and could be abused for fraud, cyberattacks, and biothreats.

Governance remains fragmented, as many countries lack the capacity to evaluate or shape advanced AI systems, leaving them dependent on technologies they cannot fully understand or control.

Additionally, existing security tools often rely on limited testing data provided by companies, according to the report.

UN Secretary-General António Guterres urged governments to act quickly. “The world cannot regulate what it does not understand. The potential is enormous, but the risks are real, and the cost of waiting is increasing,” he said in a statement.

Political and technological leaders from around the world, together with the United Nations digital technology agency, announced yesterday the creation of a new commission to address the development of artificial intelligence amid growing concerns about its possible risks.